The air-sea transfer of carbon dioxide can be viewed as a dynamical system through which atmospheric and oceanic processes push surface waters away from thermodynamic equilibrium, while diffusive gas transfer pulls them back towards the equilibrium. These push/pull processes drive extremely large sub-seasonal, seasonal, and interannual variability in air-sea carbon fluxes, the quantification of which is critical both for diagnosing the ocean response to fossil fuel emissions and natural variability, and for attempts to mitigate anthropogenic climate disruption through intentional modification of surface ocean biochemistry. We present a new autoregressive model of air-sea CO₂ exchange where these push/pull processes can include deterministic and stochastic components. The diffusive gas transfer provides the fundamental negative feedback whose strength can be estimated using the lag-autocorrelation of air-sea CO₂ flux anomaly, explaining the temporal spectrum of air-sea CO₂ exchange.  Furthermore, partial integration of the autoregressive model provides an approach for attributing air-sea carbon flux to specific mechanisms including thermodynamics, biological uptake, ocean transport terms, and atmospheric transients. We apply this new framework to model outputs and observational datasets of air-sea CO₂ fluxes. In the Southern Ocean, decomposition of seasonal air-sea carbon flux shows the dominant role of biological carbon pumps that are partially compensated by the transport convergence. We then apply the framework to analyze feedbacks to mesoscale iron and alkalinity release, explicitly quantifying transport feedbacks due to ocean currents and turbulent mixing and eventual impacts on net air-sea carbon flux. Our attribution framework provides a powerful new analytical technique for diagnosing earth system model outputs and quantifying the impacts of human perturbations to the ocean carbon cycle.